【摘要】 压电陶瓷器件这类复合材料在精密定位和微位移控制中得到广泛的应用,但其存在压电相互转化时寄生的难以建模的迟滞非线性现象的不足。通过采用一种和前向神经网络结构类似,以BackLash迟滞算子为激励函数的网络结构,并利用最小二乘法调整网络权值,从而无需考虑具体物理对象而实现对这类复杂迟滞非线性环节的建模,进而推导出所提出网络结构的逆模型。采用所建逆模型来对迟滞非线性环节进行补偿,仿真结果表明:提出的迟滞环建模方法及其控制算法可实现对迟滞环的线性补偿。更多还原

【Abstract】 Undesired complex hysteretic nonlinearities are present to a varying degree in piezoelectric sensors and actuators.This paper presents a kind of forward neural networks-like net structure.By using recursive least squares(RLS)to regulate the net weights,this net structure can model hysteresis without taking into account of the underlying physics.Further more,the paper put forward the inverse model of the net structure to compensate hysteresis.The results of simulation suggest the effect of the modeling and controlling methods.更多还原